Necessity of Prophylactic Extrapleural Chest Tube During Primary Surgical Repair of Esophageal Atresia: A Systematic Review and Meta-Analysis

Background Esophageal atresia is corrected surgically by anastomosing and recreating esophageal continuity. To allow the removal of excess fluid and air from the anastomosis, a prophylactic and temporary intraoperative chest tube (IOCT) has traditionally been placed in this area during surgery. However, whether the potential benefits of this prophylactic IOCT overweigh the potential harms is unclear. Objective To assess the benefits and harms of using a prophylactic IOCT during primary surgical repair of esophageal atresia. Data Sources We conducted a systematic review with a meta-analysis. We searched Cochrane Central Register of Controlled Trials (2021, Issue 12), MEDLINE Ovid, Embase Ovid, CINAHL, and Science Citation Index Expanded and Conference Proceedings Citation Index—(Web of Science). Search was performed from inception until December 3rd, 2021. Study Selection Randomized clinical trials (RCT) assessing the effect of a prophylactic IOCT during primary surgical repair of esophageal atresia and observational studies identified during our searches for RCT. Data Extraction and Synthesis Two independent reviewers screened studies and performed data extraction. The certainty of the evidence was assessed by GRADE and ROBINS-I. PROSPERO Registration A protocol for this review has been registered on PROSPERO (CRD42021257834). Results We included three RCTs randomizing 162 neonates, all at overall “some risk of bias.” The studies compared the placement of an IOCT vs. none. The meta-analysis did not identify any significant effect of profylacitic IOCT, as confidence intervals were compatible with no effect, but the analyses suggests that the placement of an IOCT might lead to an increase in all-cause mortality (RR 1.66, 95% CI 0.76–3.65; three trials), serious adverse events (RR 1.08, 95% CI 0.58–2.00; three trials), intervention-requiring pneumothorax (RR 1.65, 95% CI 0.28–9.50; two trials), and anastomosis leakage (RR 1.66, 95% CI 0.63–4.40). None of our included studies assessed esophageal stricture or pain. Certainty of evidence was very low for all outcomes. Conclusions Evidence from RCTs does not support the routine use of a prophylactic IOCT during primary surgical repair of esophageal atresia.


INTRODUCTION
Esophageal atresia refers to a group of congenital anomalies in which the continuity of the esophagus is interrupted (1). Tracheoesophageal anomalies are divided into subtypes depending on anatomy and the most prominent (85%) subtype has a tracheoesophageal fistula to the distal esophageal segment (1). The prevalence of esophageal atresia varies according to country and time period (2)(3)(4)(5)(6)(7)(8). Observational studies from 1981 to 2018 have estimated the prevalence to span from 0.88 to 4.55 per 10,000 births in China and in Germany, respectively. Recent European studies suggest that the prevalence is relatively stable over time (9)(10)(11) and that males are most affected with a male:female ratio of 1:0.74 (9).
At birth, the neonate presents with typical drooling of saliva, inability to swallow, choking, coughing, cyanotic attacks, and distended abdomen if the subtype involves a fistula to the trachea (11). The diagnosis is confirmed by the inability to pass a feeding tube into the stomach (11,12) and a plain X-ray showing the nonprogression of the feeding tube located in a blind-ending pouch (11,12). Prenatal diagnostics having improved from 26 to 36% during the last 30 years (9). Postnatal diagnosis occurs on the first day after birth in 83% of cases, the remaining 15% of cases within the first week, and only in 1.2% after the first week (9).
Left untreated, the condition is fatal due to starvation, infection, and respiratory complications and survival therefore relies on early surgical correction (1,11). The surgery aims to reconstruct the continuity of the esophagus and eliminate any possible fistulae (11,14), which can be done either as a transpleural thorascopical procedure, or as open surgery, most commonly extrapleural (11,22).
The mortality rate in isolated esophageal atresia range from 4.3 to 8.1% (7,8,17,(23)(24)(25), but varies with the type of atresia (with higher mortality and morbidity in the long gap esophageal atresia presentation), and mortality increases furthermore in case of prematurity and/or low birth weight, and with the presence of associated abnormalities, notably major cardiac and chromosomal anomalies (2,6,9,26,27). Even after hospital discharge, the children have increased mortality with postdischarge mortality is primarily due to respiratory compromise, including sudden infant death, aspiration, tracheomalacia, and reactive airway disease (23,28,29).
During the esophageal repair a prophylactic intraoperative chest tube (IOCT) has traditionally been placed close to the anastomosis to drain access fluid and air through a one-way system (51,52). The routine use of prophylactic IOCTs is now debated (11,39,40,52) and at the European Reference Network for rare Inherited and Congenital Anomalies (ERNICA) consensus conference, no consensus was found with only 21.4% of the members voting for the use of IOCTs (22). However, IOCTs are still used and reported as common as in 54% of the cases in the UK (53), 57% in Belgium (54), and 69% in an international survey, respectively (55).
IOCTs are not without drawback as they can cause insertion site infection (56) and when improperly placed, the tube can cause disruption of the site of anastomosis or penetration of proximal myotomy (57). IOCTs may also cause considerable postoperative pain, which would decrease inspiratory effort tand need for administration of more opoids, both leading to secondary effects such as atelectasis and pneumonia (58). Importantly, in some cases, IOCTs are insufficient to drain major leaks, necessitating the placement of a new chest tube (39,40).
Whether the potential benefits of the prophylactic IOCT overweigh the potential harms is therefore unclear (22). Accordingly, the objective of this study was to examine the benefits and harms of prophylactic IOCT during primary surgical repair of esophageal atresia.

METHODS
We conducted a systematic review of the existing literature according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines (PRISMA) and the Cochrane Handbook for Systematic Reviews of Intervention (59,60). The predefined methodology, and method for this review in general, is described in our protocol, registered in June 2021 (61).

Eligibility Criteria
We searched for RCTs assessing the effect of a prophylactic IOCT during primary surgical repair of esophageal atresia and related observational studies identified during our searches for RCTs.

Search and Study Selection
We searched the Cochrane Central Register of Controlled Trials, MEDLINE Ovid, Embase Ovid, CINAHL, and Science Citation Index Expanded and Conference Proceedings Citation Index-(Web of Science). The search strategy was developed by an information specialist from the Cochrane Hepato-Biliary Group. The search strategy can be found in Appendix 1.
Studies were included irrespectively of publication type, publications status, and language. Two independent reviewers (MRL and SKK) screened and found relevant studies, performed data-extraction using an EXCEL data extraction sheet, and systematically checked risks of bias. We planned to contact trial authors if relevant data were unclear or missing. A description of the data collection process can be found in our protocol.

Risk of Bias Assessment
We followed the Cochrane Handbook for Systematic Reviews of Interventions to examine the risk of bias (62), including the ROBIN-I tool for non-randomized studies (63). Two authors, MRL and SKK, independently assessed the risk of bias in the included trials. In case of disagreements, a third author (ULT) would arbitrate.

Data Synthesis
We pooled the data from relevant studies that were estimated to be clinically homogeneous using the Review Manager 5.4.1 software. If more than one study provides usable data in any single comparison, we performed a meta-analysis. We used RR for dichotomous outcomes, and by utilizing the fixed-effect (Mantel-Haenszel model).

RESULTS
A systematic search done December 3rd, 2021, identified a total of 953 records from databases and registers. A total of 894 were excluded based on the title and abstract. We assessed 19 fulltext original articles, of which following studies were included: three RCTs (64-66) and two case-control studies (51,52) for narrative description in the discussion. See Figure 1: PRISMA flowchart and

Included Trials
We identified and included three RCTs randomizing a total of 162 neonates with esophageal atresia and distal tracheoesophageal fistula into intervention and control group. The trials compared mortality, serious adverse events, intervention-requiring pneumothorax, and anastomosis leakage (see Table 2: summary of findings). The risk of bias assessment is shown in Figure 2. All trials were assessed to be at overall "some concerns" for risk of bias. None of the trials adequately describe the randomization process or referred to a publish protocol.

All-Cause Mortality
A meta-analysis of three trials, randomizing 162 participants, showed that an IOCT might result in an increased risk of mortality compared to neonates undergoing surgery without an IOCT, but the confidence interval was compatible with no effect [RR 1.66, 95% CI 0.76-3.65; P = 0.21; 2.8% of optimal information size (OIS); very low certainty of evidence; Figure 3].

Proportion of Participants With One or More Serious Adverse Events
A meta-analysis of three trials, randomizing 162 participants, showed that an IOCT might result in an increased risk of having a serious adverse event compared with neonates with esophageal atresia undergoing surgery without an IOCT, but the confidence interval was compatible with no effect (RR 1.08, 95% CI 0.58-2.00; P = 0.81; 7.4% of OIS; very low certainty of evidence; Figure 4).
The serious adverse effects assessed in the trials were respiratory complications including respiratory distress, pneumonia, pneumothorax, lung collapse, and apnea as well as mortality.

Proportion of Participants With an Intervention-Requiring Pneumothorax
A meta-analysis of the two trials, randomizing 112 participants, showed that an IOCT might result in an increased risk of having an intervention-requiring pneumothorax compared with neonates with esophageal atresia undergoing surgery without an IOCT, but the confidence interval was compatible with no effect  (RR 1.65, 95% CI 0.28-9.50; P = 0.58; 0.46% of OIS; very low certainty of evidence; Figure 5).

Secondary Outcomes Participants With Sepsis or Mediastinitis
One included trial (65), reporting sepsis, showed that an IOCT might result in an increased risk of having sepsis compared with neonates with esophageal atresia undergoing surgery without an IOCT, but the confidence interval was compatible with no effect (RR 3.00, 95% CI 0.14-64.26).

Participants With Anastomosis Leakage
Three trials, randomizing 162 participants, showed that an IOCT might result in an increased risk of anastomosis leakages compared with neonates with esophageal atresia undergoing surgery without an IOCT, but the confidence interval was compatible with no effect (RR 1.66, 95% CI 0.63-4.40; P = 0.30; 2.24 % of OIS; very low certainty of evidence; Figure 6).

Participants With Esophageal Stricture
None of the included trials reported on esophageal stricture.

Pain (Measured by Any Valid Score)
None of the included studies did a measurement of pain.

Summary of Main Findings
We identified and included three RCTs randomizing a total of 162 neonates with esophageal atresia and distal tracheoesophageal fistula into intervention and control group. The trials compared mortality, serious adverse events, intervention-requiring pneumothorax, and anastomosis leakage. We found no evidence of a beneficial effect of placing a prophylactic IOCT during primary surgical repair from neither of the included studies. The evidence from RCTs shows potential harm when assessing all-cause mortality and serious adverse events, but the results were very uncertain. All studies were assessed to be at overall "some concerns" for risk of bias. The risk of bias assessment is shown in Figure 2. The statistical heterogeneity was low for all our meta-analyses. It was not possible to assess the preplanned subgroups regarding esophageal stricture and pain due to the lack of relevant data.
Two observational studies (51,52) seem to support the overall results from the three RCTs in terms of mortality, serious adverse events, and anastomosis leakage that found no beneficial effect of placing a prophylactic IOCT. Furthermore, observational data from Nquyen et al. (51) suggest that the placement of a prophylactic IOCT may increase the risk of various complications such as an increase in the risk of developing esophageal stricture. These observational studies were assessed by ROBINS-I to be at overall serious (51) and critical risk of bias (52) and should therefore be interpreted with caution. Finally, we identified but excluded for various reasons (see Table 1) an additional 4 studies, no of which were in favor of routine ICOT (see Table 1).

Strengths and Limitations
This review draws strengths from the strict methodology, including following a protocol registered before the literature search began, systemically assessing for risk of bias, and adhering to all recommendations from the Cochrane Collaboration, including the use of ROBINS-I. The search strategy was developed by an information specialist from the Cochrane Hepato-Biliary Group. Our study also differs from a recent review by Anand et al. (79) on the topic by adding GRADE assessment of the included studies and abstaining from mixing RCTs with observational studies in the meta-analyses. In Anand et al., the meta-analysis included a mix of extrapleural and transpleural repair (51,52) and an observation study, where some of the participants received IOCTs by a non-prophylactic indication (74). Although the overall conclusions in the present study are fairly similar to the study by Anand et al., inclusion of nonrandomized studies with their inherently different study designs in a meta-analysis may severely compromise the validity of their results, as their lack of randomization makes them highly at risk for confounding bias resulting in an imbalance in prognostic factors associated with the outcome (80). We only identified three RCTs, systematically comparing the intervention with an IOCT to no IOCT in 162 neonates undergoing primary repair for esophageal atresia. None of our meta-analyses reached the optimal information size. In addition to evaluate overall improvement in treatment techniques and clinical outcomes, future trials should also assess pain and esophageal stricture as this would be an important outcome for the children and parents. Importantly, the associated malformations and genetic aberrations often found in esophageal atresia and the difference in exact anatomical presentation [with or without fistula(e)] make this a relatively heterogeneous pathology. The patients included in these RCTs all presented with distal tracheoesophageal fistula, but with various details on pre-surgical gap length and on associated malformations; further complicating the comparison between studies and the transferability of the conclusion to other patient subgroups.
Esophageal atresia is a relatively rare condition; particularly considering the numerous subtypes with various possible    anatomical presentations and associated malformation. Rare diseases pose challenges to methodology when designing RCTs that are adequately powered to draw definitive conclusions, as small patient sample sizes are statistically vulnerable to small deviations in the observed number of outcomes (81). Innovative clinical trial methods minimizing sample size requirements (82) and optimal research infrastructure (83), possibly through international collaborations, may improve future productivity of robust research in esophageal atresia.

CONCLUSION
We did not identify any studies advocating for the use of prophylactic IOCTs. Based on the limited amount of research on this topic and results from the included studies, we did not find sufficient evidence to support or discontinue the routine use of prophylactic IOCTs for neonates undergoing surgical repair of esophageal atresia, as all confidence intervals were compatible with no effect. Further trials, ideally multicentric, are warranted to explore the effects of the prophylactic IOCT for neonates undergoing surgical repair of esophageal atresia. Importantly, future trials should adhere to SPIRIT guidelines (84).

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

AUTHOR CONTRIBUTIONS
ML drafted the protocol and review, extracted data, coordinated the review, analyzed the data, and revised the review. SK drafted and revised the protocol, extracted data, analyzed the data, interpreted the data, commented on, revised the review, interpreted the data, provided supervision, and provided a methodological and statistical expertise. SH revised and commented on the protocol and review and provided supervision. JO and MF revised and commented on the protocol and review and provided clinical expertise. SP revised and commented on the review. UL-T drafted the protocol, conceived and designed the review, revised the protocol, commented on, revised the review, provided supervision, and clinical expertise. All authors contributed to the article and approved the submitted version.

ACKNOWLEDGMENTS
The review author team would like to acknowledge the information specialist Sarah Klingenberg from the Cochrane Hepato-Biliary Group for the development of our search strategy.